home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!auspex-gw!guy
- From: guy@Auspex.COM (Guy Harris)
- Newsgroups: comp.sys.sun.misc
- Subject: Re: Solaris 2 vs Windows NT: Sun is losing the desktop on price
- Message-ID: <16584@auspex-gw.auspex.com>
- Date: 22 Jan 93 21:42:21 GMT
- References: <1jf1l0INN6ak@gap.caltech.edu> <1jo3c6INNl95@darkstar.UCSC.EDU> <16570@auspex-gw.auspex.com> <1johm0INNp4e@darkstar.UCSC.EDU>
- Sender: news@auspex-gw.auspex.com
- Organization: Auspex Systems, Santa Clara
- Lines: 25
- Nntp-Posting-Host: auspex.auspex.com
-
- >I thought ALL sun4 line HAD to have FPUs, and that no-fpu stuff was
- >only forthe sun3 line.
-
- Nope. Some 4/110s were sold without an FPU, and, for a while, the 4/280
- I had at Sun didn't have an FPU, becasue the FPU it came with was
- defective.
-
- The Sun-3s had several different compiler options for no-FPU machines,
- machines with a 68881, machines with the Sun FPA, and for programs that
- should figure out at run time which option they had.
-
- For SPARC, Sun took the approach that:
-
- 1) all floating-point hardware should implement (mostly) the
- same instruction set (a few instructions have been added, and
- some have disappeared, over time, although I don't know
- whether any other than the old 96-bit extended precision
- ones, which Sun's software didn't use, disappeared);
-
- 2) on machines that don't have the hardware, the software must
- catch the illegal instruction traps caused by attempts to
- execute the floating-point instructions, and must emulate
- them in software - the SunOS kernel does that;
-
- so there isn't a "software floating point" compiler option.
-